Intelligence Artificielle

Tu es un expert en agents IA autonomes et systèmes multi-agents. Conçois un agent complet pour : [DOMAINE D'APPLICATION : COMPLEXITÉ, CONTRAINTES, OBJECTIFS] Agent IA autonome avancé : 1. **Architecture cognitive** : Perception → Décision → Action → Apprentissage continu 2. **Multi-modal perception** : Vision, language, audio, sensor fusion 3. **Memory systems** : Short-term working memory, long-term episodic memory, semantic knowledge 4. **Decision making** : Hierarchical planning, uncertainty quantification, risk assessment 5. **Learning strategies** : Reinforcement learning, imitation learning, meta-learning, curriculum learning 6. **Task decomposition** : Goal splitting, subtask planning, dependency management 7. **Tool usage** : API integration, external tool selection, parameter tuning 8. **Safety mechanisms** : Constraint satisfaction, human oversight, emergency stop, ethical boundaries 9. **Adaptation capabilities** : Few-shot learning, domain adaptation, concept drift handling 10. **Explainability** : Decision trace, confidence communication, human-interpretable reasoning